OCCASIONAL PAPERS cor/r l/brauy i.'UV 2 T •92 ° f the HARVARC MUSEUM OF NATURAL HISTORY The University of Kansas Lawrence, Kansas NUMBER 106, PAGES 1-37 NOVEMBER 10, 19S3 A "CURSORIAL” CROCODILIAN FROM THE TRIASSIC OF LESOTHO (BASUTOLAND), SOUTHERN AFRICA BY K. N. Whetstone 1 and P. J. Whybrow 2 For many years most vertebrate paleontologists regarded the crocodilians as a stable side-branch of the archosaurian reptiles, of interest because of their supposedly primitive structure and their relationship (albeit “distant”) to the dinosaurs and birds. Recent studies by A. D. Walker (1968, 1970, 1972, 1974) and E. Buffetaut (1979) have rekindled interest in the crocodilians as a morpho- logically diverse group with terrestrial, marine and amphibious specializations. The extent of this diversity is illustrated by Walker’s arguments (1972, 1974) that early crocodilians were close to the basal stock of birds. Prior to this report, our knowledge of the earliest (Triassic) crocodilians was based upon at least fifteen published specimens, assigned to six genera. While this fossil record is by no means meager, it is very poorly known, particularly in the anatomical details which are important in phylogenetic and functional studies. This lack of knowledge is partly the result of the poor preservation of the material and the difficulties of preparing it from indurate sandstones and ironstones. In 1966 and 1967 a joint British Museum (Natural History) — London University expedition recovered a partial crocodilian skele- ton from Upper Triassic sediments in Lesotho, southern Africa. The fossil is of a lightly built individual with a suite of adaptations for rapid terrestrial locomotion. As a result of its detailed preserva- 1 University of Kansas and Museum of Natural History, Lawrence, Kansas; present address: Union Texas Petroleum, 2500 First Oklahoma Tower, Okla- homa City, Oklahoma. 2 British Museum (Natural History), London, England. 2 OCCASIONAL PAPERS MUSEUM OF NATURAL HISTORY tion, it is possible to describe many features that are strikingly different from those of Recent crocodilians. These characters pro- vide new support for Walkers hypothesis of a close relationship between birds and crocodilians and further expand our knowledge of the morphological diversity of the Crocodylia. The new Lesotho specimen is distinct from the genus Orthosuchus and is assigned to the new genus and species, Lcsothosuchus charigi. The type specimen was collected and prepared by P. J. Why- brow, whose contribution is acknowledged by the joint authorship of this paper. The text which follows is by the senior author, who takes full responsibility for the interpretations presented. PREVIOUS STUDY AND CLASSIFICATION OF TRIASSIC CROCODYLIA The first Triassic reptile to be correctly recognized as a croco- dilian was Notochampsa istcdana from Cape Province, South Africa (Broom, 1904). The type material of this species has never been prepared from the matrix. As exposed, it consists of an impression of the nasal and temporal regions of the skull (Fig. 1A), part of the right mandible, a right pectoral girdle with the glenoid partly covered, the proximal end of the right humerus, parts of the left forelimb, parts of the left hindlimb (figured by von Iluene, 1925), and impressions of the dorsal armor. Notochampsa longipes, de- scribed by Broom (1904) in the same work, was later selected by Ilaughton (1924) as the type species of his genus, Erythrochampsa. The type specimen has been figured by Nash (1971). It consists of dorsal and ventral osteoderms, both ischia, a right pubis, a possi- ble left pubis, the metatarsus, a tibia, and a fibula. Von Iluene (1925) identified a radius, ulna, and femur in the block, but these are not evident in Nash’s photographs. An additional, fragmentary specimen was referred to this species by Broili and Schroeder (1936). Although the genera Notochampsa and Erythrochampsa can be confidently assigned to the Crocodylia, their type material is indeterminate. Owing to their poor preservation and description, it would be possible to assign any of the other Triassic forms to those taxa by attributing the few discernable differences to inade- quate representation and fault}' preservation. For the present, both Notochampsa and Erythrochampsa are best considered nomina dubia. A single specimen of a crocodilian from Triassic sandstones of the Connecticut valley, named longipes by Emerson and Loomis (1904), was referred bv them to Stcgomus but was later given its own genus, Stcgomosiiclws , by von Iluene (1922). It, too, is poorly preserved, although a good impression of the skull roof, dorsal armor, and metatarsals can be found on the counterslab. TRIASSIC CROCODILIAN FROM SOUTHERN AFRICA 3 Walker (1970) argued its crocodilian affinities and attempted a cranial reconstruction based upon a cast of the counterslab. I have examined this cast and agree that Stegomosuchus is a crocodilian. An expanded squamosal with a longitudinal ridge and a flange on the lateral margin are distinctive crocodilian features. There also appears to be a posterior palpebral element preserved on the right postorbital. Unfortunately, the preservation is too incomplete to determine other details of the skull roof with any confidence. An outline of the limbs is given by Emerson and Loomis (1904), but only the metatarsals appear as relief on the casts. Lacking details of skull construction, limb girdles, appendages, and vertebral col- umn, Stegomosuchus can only be classified as a crocodilian of un- determined affinity, probably of protosuchian grade. The first Triassic crocodilian to be represented and described in detail was Protosuchus richardsoni (Fig. IB); first named by Brown (1933, 1934) and later monographed by Colbert and Mook (1951). Most of the skeleton is preserved in the type (AMNH 3024), although there are several other specimens, all from the Dinosaur Canyon Sandstone in Arizona, U.S.A. Also represented by good material is Orthosuchus stormhergi (Fig. 1C), known by two specimens from Lesotho which were described by Nash ( 1968, 1971, 1975). As originally restored by Colbert and Mook, Proto- suchus differed substantially from Orthosuchus in having no antor- bital fenestra, the parietal excluding the frontals from the upper temporal fenestrae, the dentarv notched posteriorly and the quad- rate unfenestrated. These differences at first seemed sufficient to place them in separate families, but a new specimen of Protosuchus described by Crompton and Smith (1980) resolved some of the discrepancies. Protosuchus is now known to have an antorbital fenestra and a pneumatic quadrate. In the classification that fol- lows. I have placed Orthosuchus and Protosuchus in the same fam- ily, Protosuchidae. The new Lesotho specimen is distinct from Orthosuchus at the generic level, but is placed with it in the new subfamily, Ortho- suchinae for reasons discussed below. In previous studies by Nash (1971. 1975) and Whetstone and Martin (1979), the Lesotho speci- men, heretofore unnamed, was referred to the genus Notochampsa. Walker (1972) mentioned the specimen by number only. Lesotho- suchus and Orthosuchus were small animals with more or less gracile build. Terminal length for both taxa was undoubtedly less than one meter. Eopneumatosuchus colhcrti is a crocodilian recently described by Crompton and Smith (1980). It is known only by a skull col- lected from sediments of probable Upper Triassic age in Arizona. It is characterized by the highly pneumatic construction of the 4 OCCASIONAL PAPERS MUSEUM OF NATURAL HISTORY braincase and an elongate laterosphenoid. I have placed it in its own subfamily of the Protosuchidae. Numerous other taxa liave been associated with protosuchian crocodilians at one time or another. These include Dyoplax , Erpeto- sucliiis , Hesperosuchus , Proterochampsa , Saltoposuchus , Pedetico- sciiirns , Microchampsa , Platyognathus , Sphenosiichus , Pseudhespero- su chits , Hemiprotosuchus , and an unnamed form from Welsh fissure deposits (Kermack, 1956; Crush, MS.). Of these, Krebs (1976) considered Dyoplax , Erpetosuchus , Hesperosuchus , Protero- champsa, , and Saltoposuchus to be thecodontians of little or no relation to the Crocodylia. Pedetlcosaurus and Microchampsa are based on insufficient material to be confidently assigned to any particular archosaur group. Platyognathiis is known principally from a referred specimen (Simmons, 1965), which is much too fragmentary to support Simmons’ conclusion that it had a meso- suchian palate. The coracoid does appear to be elongated, how- ever, and this might indicate crocodilian relationships. The teeth have expanded roots, but are serrated. Until more complete material is known, the best assignment for Platyognathus is “PCroco- dylia.” The remaining taxa, Sphenosiichus , Pseudhesperosuchus , Hemi- protosuchus and the unnamed Welsh genus can be loosely grouped into the Sphenosuchidae. As discussed below, they share some diagnostic features with “true” crocodilians, but are very primitive in other respects. To include them in the Crocodylia would greatly modify traditional interpretations of that group, diminish the diag- nosis, and possibly create a grade taxon from a demonstrably monophyletic (sensu Hennig, 1966) one. On the other hand, it seems advisable to remove them from the Thecodontia. In informal nomenclature I have followed Walker (1970) in using the name, croeodylomorphs to designate the true crocodilians plus the Spheno- suchidae. In the formal classification which follows, the spheno- suchids appear as the “sister-group” to the crocodilians. All of the sphenosuchids (with the possible exception of Hemiprotosuchus) have an unusual coracoid morphology which can either be inter- preted as uniquely derived for the group or primitive for the croco- dvlomorphs. This feature may link them to Hesperosuchus if Colbert’s (1952) “problematical bone” of Hesperosuchus is really a coracoid. The following classification is used in this study. Despite the unsettled nature of archosaur taxonomy and the reluctance of many workers to include birds within the Archosauria, I have re- jected a traditional “grade” classification of archosaurs. Taxa which I believe to be paraphyletic grades are indicated by quotation marks and the plesion convention of Patterson and Rosen (1977) is used to eliminate an unnecessary higher taxon. TRIASSIC CROCODILIAN FROM SOUTHERN AFRICA 5 Division Archosauria Subdivision Nidosuchia, new taxon Plosion Sphenosuchidae: Sphcnosuchus , Pscudhcspcrosuchus, Welsh fissure deposit crocodylomorph, ?Hcspcrosuchm, PHemiprotosuchus Infradivision Crocodylia Crocodylia, insertae sedis (“Protosuchia”) Family Protosuchidae: Subfamily Protosuchinae: Protosuchus Subfamily Eopneumatosuchinae: Eopneumatosuchus Subfamily Orthosuchinae: Orthosuchus, Lesothosuchus “Protosuchia” indet.: Stegomosuchus Notochampsa (nomen dubium), Enjthrochampsa (nomen dubium) Order Mesoeucrocodylia, new taxon Mesoeucrocodylia, insertae sedis ( “Mesosuchia” ) Suborder Eusuchia Infradivision Aves TAXONOMY AND MATERIAL Division Archosauria Infradivision Crocodylia “Protosuchia” Family Protosuchidae Lesothosuchus new genus Lesothosuchus charigi new species Diagnosis: Small crocodilian with relatively long limbs; postorbital with a facet for a supraorbital (palpebral) bone; temporal fenestrae moderate in size and pear-shaped; squamosal broad, with a longi- tudinal ridge near the lateral margin; paroccipital process centrally placed, relatively short and broad; carotid artery enclosed by bone posterior to the otic region; squamosal overhangs the tympanic cavity; braincase sutures to the quadrate anteriorly, retains a cotyle posteriorly; quadrate hollow and highly fenestrated; periotic pneu- maticity developed; floccular recess present within endocranium; lower temporal fenestra reduced; vertebral centra platycoclous ex- cept the posterior caudals; coracoid elongate, expanded sternally, and “waisted”; humerus with an expanded, convex head; ulna with two proximal cotyles and a rounded, convex articulation for the radius; dorsal armor finely sculptured. Derivatio nominis: generic name recognizes the provenance of the type specimen; species name in honor of A. J. Chari g who has contributed much to our knowledge of early archosaurs. Type: BMNH RS503 (Table 1); left otic and temporal region of skull; left dentary; one cervical, two dorsal, and two caudal verte- brae; a caudal chevron; some rib fragments; dorsal end of the right 6 OCCASIONAL PAPERS MUSEUM OF NATURAL HISTORY scapula; right coracoid; right and left humeri; a left ulna; most of left femur; distal end of right femur; a right fibula; right and left metatarsals of the second digit; two phalanges from the pes; several osteoderms. Type locality: Lithipeng South, Mohales Hoek district, Lesotho; from northwest face of vertical slope at longitude 24°44' and lati- tude 30°19'S. Horizon: Red Beds of Stormberg Series, just below Cave Sand- stone; photograph in Charig (1969) shows stratigraphy of the region. COMPARATIVE DESCRIPTION Cranium The left dentary and most of the left half of the posterior por- tion of the skull are preserved. This includes the skull roof, occiput, otic capsule, parocciput, the dorsal surface of the quadrate and part of the postorbital bar. The basieranium is missing. The dentary (Fig. 6A,B) is a thin, slender bone with alveoli for eight or nine teeth preserved. The bony tooth sockets are completely separated by bone to the rear of the tooth row. In Recent erocodilians the posterior mandibular teeth of juveniles are set in a groove. As the individual grows the groove is progressively divided into alveoli from the front to the rear. The dentary be- Table 1. — Measurements in mm for Lesothosuchus charigi (BM (Nil) R 8503). Temporal width Width across quadrates Skull 48* 60° Anterior height Anterior Anterior to top of width height Length neural canal Cervical centrum 6.1 7.5° 7.2 11.5 Mid-dorsal 6.3 6.3 9.8 9.7 Anterior caudal 6.3 6.7 9.6 9.3 Posterior caudal 4.7 4.7 10.2 ? Max prox Max dist Proximal Distal Length width width thickness thickness Humerus 56.7 12.6 11.1 5.9 6.6 Ulna 51.3 9.1 6.5 5.1 4.0 Femur 70 est.* ? 15.1 ? 10.2 Fibula 66.2 8.0 7.2 2.7 4.7° * Skull widths are doubled measurements to midline; anterior cervical measure- ments include keel; femur length is conservatively estimated by reconstruc- tion of the proximal end based upon comparison with Triassic and Recent taxa; distal fibula may be somewhat crushed. TRIASSIC CROCODILIAN FROM SOUTHERN AFRICA comes fully septate only in specimens which are near “terminal” size. Fully septate dentaries are found in the type specimens of both Lesothosuchus and Orthosuchus. If Triassic crocodilians de- veloped as Recent ones do, this indicates that the type specimens are adults, that their small size cannot be attributed to youth, and that the differences in skeletal proportions between the types do not represent ontogenetic variation in a single species. There are two grooves on the internal surface of the dentary, just ventral to the tooth row. The more dorsal of these becomes en- closed anteriorly to form the alveolar canal. The ventral groove held the anterior extension of the 1 angular bone which formed the floor of Meckel’s canal. Behind the ventral groove at the postero- ventral corner of the dentarv is a rounded surface for articulation with the angular. Although there is no broad, rugose articular sur- face of the sort that is found in Alligator , there probably was some lateral exposure of the angular bone in this area. The dentarv is highly sculptured anteriorly, but less so pos- teriori}’. At the posterior end, there is no notch for the mandibular foramen. This notch is also absent in Orthosuchus, but, to the best of my knowledge, in no other crocodilians. In Orthosuchus, Nash restores the mandibular foramen as being posteriorly posi- tioned and quadrangular, also unique among the Crocodylia. The skull roof of Lesothosuchus is strongly sculptured with ir- regular pits and ridges (Fig. 2A). The anterior roofing elements preserved are the dorsal portion of the postorbital (Fig. ID) and a fragment of the frontal. The postorbital bone forms the back of the orbit, part of the skull roof, and the anterior margins of the upper and lower temporal fenestrae. In eusuchian crocodilians, the postorbital is indented on its anterolateral surface so that the anterolateral corner of the temporal skull roof overhangs the cheek at the back of the orbit. This indentation is not developed in Leso- thosuchus or other protosuchian crocodilians. However, both Orthosuchus and Lesothosuchus have the postorbital indented pos- teriorly so that it forms part of the shelf which covers the otic re- gion. At the posterior corner of the orbit the postorbital has a shallow, crescent-shaped depression. In Orthosuchus this depres- sion receives the posterior palpebral, or supraorbital bone. In Re- cent crocodilians, these dermal elements are loosely attached to the cranium. They are rarely preserved in their natural position in fossils. Extant forms have only a single, anterior supraorbital attached to the prefrontal and lacrymal. Orthosuchus, Lesotho- suchus, Protosuchus (see Walker, 1970; Crompton and Smith, 19S0), and Stegomosuchus appear to have been unique in having an addi- tional posterior element. Since a posterior papebral bone occurs in all of the accepted Triassic taxa, we presume this to be the primitive 8 OCCASIONAL PAPERS MUSEUM OF NATURAL HISTORY condition in the Crocodylia. Better material of sphenosuchid skulls could serve as an “outgroup” comparison to test this polarity. Most of the preserved portion of the skull roof is formed by the squamosal bone, which extends laterally to overhang the otic region. The squamosal, postorbital, parietal, and, probably, the Figure 1. — Restorations of the dorsal aspect of the posterior skull roof in A, Xotocliainpsa istedana after von Iluene (1925); B, Protosuchus richardsoni after Colbert and Mook ( 1951); C, Orthosuchas stormbergi after Nash (1975); D, Lcsothosuchus charigi, BM (NH) R 8503. Seale is 2 cm. TRIASSIC CROCODILIAN FROM SOUTHERN AFRICA 9 frontal form the dorsal margins of the superior temporal fenestra. The fenestrae are moderate in size and pear shaped, in contrast to the large, rounded fenestrae of both specimens of Orthosuchus, and to the small, oval fenestrae of the type specimen of Noto- champsa (Fig. 1). Stegomosuchus, however, is quite similar in this region. The size, shape and position of these openings vary ontogenetically in living crocodilians (Nash, 1971), but appear to be a useful taxonomic indicator when individuals of a similar size are compared. As in all crocodilians, the parietal of Lesothosnchus is a median, unpaired element. This is in contrast to the paired parietals of typical proterosuchian and pseudosuchian archosaurs. There is no pineal opening. The posterior portions of the frontals arc probably preserved on the specimen, but I have been unable to locate a fronto-parietal suture. The parietal extends posteriorly onto the broad occipital surface of the skull (Fig. 2C). As is characteristic of the Crocodylia, the occiput slopes abruptly downward from the skull roof. The occiput is formed by the parietal, squamosal, supraoccipital and otoccipital (fused exoccipital/opisthotic) bones. The parietal contributes only a small, centrally situated process to form the top of a central crest. On either side of this crest are impressions for the insertion of cervical musculature onto the supraoccipital. The parietal and supraoccipital form the dorsal and ventral margins of the small posttemporal fenestra. This fenestra leads into a small pneumatic sinus, which, in turn, connects with the tympanic cavity and the superior temporal fenestra. In modern crocodilians the occipito- temporal artery passes along the roof of the dorsal pneumatic cavity and through the posttemporal fenestra (Hochstetter, 1906). A simi- lar course is indicated for Lesothosnchus . The squamosal extends ventrally onto the dorsolateral corner of the occiput. On the occipital surface I have been unable to trace the suture with the otoccipital beyond the lateral-most margins. On the tympanic surface the suture is squamous, extending much farther ventrally. Most of the occipital surface of Lesothosnchus is formed by the otoccipital which extends laterally and ventrally from the supra- occipital to surround most of the foramen magnum. On either side of the foramen magnum it forms a lateral (paroccipital) process, ventrally situated at about the level of the otic capsule (Fig. 3A). This process is relatively short and is preserved only as a thin sheet. Except for the area at its base, it is completely free of the overlying squamosal. This contrasts with the primitive archosaurian condition found in most theocodontians and dinosaurs, in which the paroccipital process is long and dorsally placed. The squamosal overlies it for most of its length (either arching over it to form a 10 OCCASIONAL PAPERS MUSEUM OF NATURAL HISTORY TRIASSIC CROCODILIAN FROM SOUTHERN AFRICA 11 large post-temporal fenestra or suturing along its antero-dorsal margin) and curves downward near the lateral terminus to form a socket for the dorsal knob of the quadrate. In some ways, the paroeciput of cusuchians resembles theoeodontians more closely than Lesothosuchus — the paroeciput is long, dorsally situated and underlies the squamosal, which articulates with the quadrate just anterior to the lateral terminus of the paroeciput. In Lesothosuchus the otoccipital also forms the dorsal and lat- eral margins of the foramen magnum and the lateral portion of the occipital eondvle. The central portion of the eondyle, presumably formed by the basioccipital, is not preserved in the type specimen. Immediately lateral to the eondyle is a groove leading into a dor- sally directed foramen which marks the exit of the vagus nerve from the cranium. Lateral to this are one or more additional fora- mina, which are badly crushed. One of these must be the carotid foramen since it overlies a bony tube which corresponds in posi- tion to the carotid eanal of Recent taxa. Croeodilians are unusual among diapsid reptiles in having the carotid artery enclosed by bone before it enters the ear region. The otie region of the cranium is particularly well preserved (Figs. 2B,D; 3). It is roofed by both the squamosal and postorbital. The postorbital is broad and is greatly indented behind the post- orbital bar. On its ventral surface it bears a pit for the dorsal head of the laterosphenoid. This type of laterosphenoid articulation is widely distributed among the arehosaurs, occurring in Recent eroe- odilians, ornithisehian dinosaurs, and Mesozoic hesperornithiform birds, for example. In the latter group the laterosphenoid articu- lates with the postorbital process of the frontal bone. The squamosal broadly overhangs the otie cavity, dorsal and posterior to the squamosal articulation with the quadrate bone. On the dorsal surface of the squamosal near the lateral margin is a sinuous longitudinal ridge bordered laterally by a shallow groove. In Recent croeodilians this ridge and groove represent the attach- ment for the thickened integument that extends ventro-laterally over the muscular operculum (Ohrklappe). On this basis we can infer an opercular structure for Lesothosuchus. In Recent taxa this external ear receives blood supply from a branch of the temporal artery, which exits the dorsal periotic sinus laterally ( Iloehstetter, 1906). There is some evidence for the presence of this arterial supply in Lesothosuchus , where a series of grooves radiate out- ward from a foramen near the squamosal/ prootie/ otoceipital junc- ture (the foramen is described below as the lateral opening of the dorsal periotic sinus). Lesothosuchus differs from eusuehians in Figure 2. — Lesothosuchus chart gi, BM (NH) R 8503; skull in A, dorsal; B, lateral; C, posterior; D, ventral views. Scale is 2 cm. B and D are stereo- pairs. 12 OCCASIONAL PAPERS MUSEUM OF NATURAL HISTORY lacking a thickened ventral lip for the attachment of the operculum and from Orthosuchus and eusuchians in having the lateral margin of the squamosal only weakly downturned. The quadrate is preserved in an approximately natural position. It is strongly inclined, but less so than in eusuchians. The anterior end extends far medially to suture with the squamosal, prootic, postorbital, and laterosphenoid, all of which comprise the facies articularis anterior. In this region, the quadrate of eusuchians su- tures with these same bones and also with the parietal. In Lesotho- suclius there may be some slight contact with the parietal at the rear of the superior temporal fenestra, but it would be negligible. Figure 3 . — Lcsothosuchus charigi , RM (Nil) R 8503: skull in A, lateral and B, ventral aspects; key to Figure 2 R, D. Sq, squamosal bone; Pro, prootic, Oto, otoceipital; Qu, quadrate; ant. pneu., antrum pneumatieum, fac. artic. post., facies articularis posterior. In Fig. A the quadrate is shaded; in Fig. B the pneumatic cavities and foramina are shaded. TRIASSIC CROCODILIAN FROM SOUTHERN AFRICA 13 The quadrate forms part of the lateral wall of the fenestra, curving dorsallv to be applied as a thin sheet along the medial margins of the squamosa] and postorbital. The anterior end of the quadrate is compressed dorso-ventrally, restricted to the area dorsal to the rostral periotic sinus. This is in striking contrast to the anterior process of cusuchians, which extends ventrally to contact the basi- cranium, in some cases incorporating a pneumatic sinus. Major modifications of this sort occurred throughout the otic region of eusuchians as a result of the increased area of fixation for the akinetic quadrate. Lesothosuchus resembles mcsosuchians and eusuchians in hav- ing an additional articulation for the quadrate at the rear of the tympanic cavity, the facies articularis posterior (Fig. 3). While eusuchians have a sutural contact here, Lesothosuchus retains a cotyle for the quadrate, the first record of such a primitive articu- lation in a crocodilian. The presence of a cotyle here does not in- dicate streptostyly, which is precluded by the anterior suturing of the quadrate to the braincase and skull roof. It does, however, provide an intermediate between the loosely articulated quadrates of generalized theocodontians and sphenosuchids and the fixed, extensively sutured quadrates of mesosuchians and eusuchians. The articular cotyle is supported by two processes (Fig. 3B). The posterior process originates just anterior to the paroeeipital process and is undoubtedly part of the otoccipital bone. The an- terior process is closely associated with the posterior part of the otic capsule and cochlear canal and is probably also part of the otoccipital. There is no clear separation of the prootic and otoc- cipital in this area, and it is also possible that the prootic forms part of the cotyle, as suggested by Whetstone and Martin (1981). This is the case in the crocodylomorph, Sphcnosuchus, where the remainder of the posterior articulation is of squamosal origin. Eopneumatosuchus resembles eusuchians in lacking a prootic con- tribution to the posterior articulation. It has no cotyle, but also lacks a deeply interdigiting suture. In Lesothosuchus the posterior articular area is shaped like an inverted pear. It is crushed somewhat towards the midline so that it is separated from the quadrate. The articulation is underlain and medially undercut by a pneumatic cavity in the otoccipital. The corresponding part of the quadrate is badly damaged. There was, presumably, a bony knob on the quadrate to fit the cotyle, but it is not preserved. The facies articularis posterior is more extensive in eusuchians than in the Triassic forms. The lateral part of the eusuchian occi- put has two distinct otoccipital/ quadrate articulations — a laterally- directed articulation lying beneath the foramen for the hyo- mandibular ramus of the facial nerve and the temporal artery (the 14 OCCASIONAL PAPERS MUSEUM OF NATURAL HISTORY “cranio-quadrate canal” of Iordansky, 1973), and a ventral ly-di- rected articulation on the paroccipital process. The parocciput extends dorsal and lateral to the foramen and is overlain by the squamosal. Only the ventral part of this articulation appears to be homologous with the facies articularis posterior of Lesotho- suchus. In Lesothosuchus the quadrate has a more vertical orienta- tion than in Recent taxa. It articulated with the otoccipital bone ventral to the hyomandibular ramus, which lay in an open groove over the dorsal surface of the posterior cotvlus. This is assumed to be the primitive situation for the Crocodylia. Eusuchians have extended the quadrate dorsally and ventrally relative to this, so that it contacts the lateral margins of the parocciput dorsally and the basicranium ventrally. In addition, the eusuchian quadrate has regained a posterior articulation with the squamosal, just an- terior to the lateral end of the paroccipital process. The extension of the posterior articulation in eusuchians mirrors the increased area of quadrate fixation found anteriorly, both being modifications for complete akinesis. In Lesothosuchus the body of the quadrate is hollow and is perforated by many foramina. These pneumatic foramina resemble those of Orthosuchus and differ from those of Protosuchus and eusuchians in being randomly situated all along the dorsal surface of the quadrate. Although only a small part of the ventral surface of the quadrate is preserved, it is perforated also. One foramen connects the tympanic cavity with the superior temporal fossa. There are at least three possible explanations for the unusual quadrate morphology. If the structure of the quadrate represents an elaboration of quadrate pneumaticity found in eusuchians, the quadrate of Lesothosuchus could have carried an extensive si- phonium from the tympanic cavity into the lower jaw. Alterna- tively, the cavity in the quadrate could simply be an extensive periotic sinus, an analog to the pneumatic cavities in the braincase. The first hypothesis would not account for the ventral perforation of the quadrate unless the siphonium of Lesothosuchus was in some manner connected to the eustachian system. Neither hypothesis would explain the foramen which extends into the superior tem- poral fossa. A third possibility is suggested by comparison with birds, the other surviving archosaur derivative. In the approximate position of the fenestrated quadrate of Lesothosuchus , birds have a temporal rete, an anastomosis between the stapedial artery and the lateral head vein (Saiff, 1974, 1976, 1978, in press; Crow and Crow, 1979). Relative to the condition in birds and other archosaurs, Lesotho- suchus and other crocodilians have extended their quadrates antero- dorsally so that this region is covered by the quadrate. In Recent crocodilians the stapedial artery is reduced, the temporal artery is TRIASSIC CROCODILIAN FROM SOUTHERN AFRICA 15 enlarged and has a posterior origin, and there is no temporal rete. All of these vascular changes may he the result of the separation of the tympanic recess from the orbit and temporal fenestra as a consequence of the increase in the fixation area for the quadrate. Triassic taxa had a more open tympanic cavity and may have had a rete which passed through the quadrate. This would explain the ventral perforation of the quadrate and the presence of a foramen leading to the musculature of the temporal fossa. Interpreting the quadrate in this way presumes a more primitive cranial arterial system than in extant crocodilians, a more diffuse rete than in modern birds, and a rather sinuous course for the stapedial artery (since it would have to swing ventrally and medially under the postorbital bar to reach the orbit). While this option seems to provide the best explanation of the morphology observed in Leso- thosuchus , it is not necessarily an exclusive explanation. The ex- tensive hollow cavity indicates some degree of quadrate pneu- maticity and there is good reason to believe that the siphon ial sys- tem is an ancestral character in crocodilians. Associated with the tympanic cavity of Lesothosuchus are three periotic pneumatic cavities. The first of these, lying just dorsal to the posterior quadrate articulation, begins in a deep recess that is an extension of the middle ear cavity. The recess leads medially into two pneumatic foramina. The ventral foramen appears to be directed anteriorly to connect with an anterior sinus, but it has not been adequately prepared. The dorsal foramen is situated at or near the junction of the squamosal, prootic and otoccipital bones within the tympanic cavity (sutures are obscured in this area). It has been prepared to an internal pneumatic sinus at the medial junction of these same bones. Via the sinus, the tympanic cavity is connected to the posttemporal fenestra posteriorly and the su- perior temporal fenestra anteriorly. The sinus corresponds to the dorsal periotic sinus of eusuehians, the antrum pneumaticum dor- sale. In eusuehians that sinus is partially divided by a thin shelf of prootic bone so that the dorso-lateral portion, carrying the oc- e : pito-temporal and orbito-temporal arteries, is separated. This dorsal portion (“post-quadrate canal” of Walker, 1972) is weakly, if at all, pneumatic and has a lateral foramen at the junction of the squamosal, prootic, and otoccipital. The ventral part (“mastoid antrum” of Walker, 1972) is extensively pneumatic (much more than in Lesothosuchus) and has a separate lateral foramen at the junction of the prootic and otoccipital. Lesothosuchus and Spheno- suchus have only a single pneumatic sinus above the facies articu- laris posterior and the divided cusuehian sinus is obviously derived from it. Immediately ventral to the posterior eotvle is a deep pneumatic recess in the otoccipital which undercuts the articular facet. This 16 OCCASIONAL PAPERS MUSEUM OF NATURAL HISTORY sinus, the antrum pneumaticnm centrale (“caudale” of some au- thors) also occurs in cusuchians, where it is restricted to the otoc- cipital just below its articulation with the quadrate. In cusuchians it communicates with the dorsal sinus by a canal medial to the pos- terior quadrate articulation. In Lcsothosuchus the canal communi- cates with the medial recess which leads into the dorsal sinus. A third periotic sinus, the antrum pneumaticnm rostrale, is formed anterior to the foramen for the facial nerve (VII). It is a deep recess of the braincase wall which joins the tympanic cavity just dorsal to the foramen. Its lateral wall appears to be formed by the quadrate, but this area is incompletely preserved. Within the sinus are fine struts of bone between the braincase and quadrate. In Eopneumatosuchus the sinus is very highly strutted. It lies be- tween the facial foramen and the trigeminal foramen, a position which can be safely inferred in Lcsothosuchus. However, in Eopneumatosuchus the sinus lies under a lateral shelf of the prootic, which articulates with the quadrate. There is a similar shelf in Sphcnosuchus, although the recess is only weakly developed. The formation of the lateral wall of the sinus by the prootic may be the primitive condition in crocodilians. In Recent taxa the corre- sponding area is pneumatic, but not obviously so since it is un- strutted. It may be a prolongation of the middle ear into a gap between the anterior arm of the quadrate and the braincase wall, or it may be incorporated into the quadrate. In either case, it over- lies the eustachian system and the cavities in the basisphenoid which surround the cerebral carotid. The Lcsothosuchus specimen has the otic capsule lying medial to the facies articularis posterior. The endocranial component of the capsule is a swollen protrusion into the endocranial cavity. Internally, there is a distinct suture separating the prootic and otoccipital (opisthotic) parts of the capsule. Above this suture is the foramen for the endolymphatic duct. The supraoccipital (epi- otic) suture is not visible. The dorsal part of the endocranial cap- sule is constricted by the expansion for the cerebrum anteriorly and by the deep floccular recess posteriorly. Recent cusuchians differ from most higher archosaurs in lacking a floccular recess. Although the endocranial cavity of Lcsothosuchus seems small in comparison with Recent taxa, it is difficult to judge the cranial volume by the small amount of surface preserved. Anterior to the capsule is the foramen nervi facialis (VII), opening laterally into the rear of the rostral periotic sinus. Anterior and dorsal to the otic capsule and near the dorsal periotic sinus is a crack which Walker (1972) interpreted as the prootic/ opisthotic suture. This interpretation is not possible since the crack runs anterior to the otic capsule. The lateral part of the capsular region is not well preserved TRIASSIC CROCODILIAN FROM SOUTHERN AFRICA 17 and mechanical preparation of the remaining matrix does not seem possible. The cochlea is completely enclosed by bone laterally, so that the fenestra ovalis can only be reached from above. This mor- phology requires a dorsally oriented stapes and oblique tympanum similar to those of Recent eusuchians. Nash (1975) restored Ortho- sucluis with a small, laterally oriented tympanum in the notch at the posterior corner of the skull roof. In light of the similarity of the tympanic region to Lesothosuchus , and the fact that the tym- panum of Recent forms always covers the dorsal pneumatic fora- mina in the quadrate, an oblique orientation seems likely for Orthosuchus as well. In Recent crocodilians the tympanum lies at an oblique angle over the inclined quadrate, extending laterally about one-third of the quadrate's length. It is completely enclosed by bone, but is separated from the squamosal at its posteromedial corner by connective and vascular tissues. In Lesothosuchus the orientation was presumably similar, but must have been more ex- tensive, laterally, to cover the dorsal foramina in the quadrate. One of these foramina is just anterior to the mandibular articulation. The posterior margin of the tympanum would have been on the quadrate and along soft tissues passing dorsal to the facies articu- laris posterior. There was probably some attachment to the squa- mosal on a crest at the rear of the tympanic cavity. At the anterior end of the tympanic region, part of the anterior wall of the quadrate is preserved. In a bit of matrix adjacent to this there is an elongate piece of bone which may be part of the quadratojugal, otherwise not preserved. A quadratojugal structure like that of Orthosuchus is likely for Lesothosuchus. Because of the proximity of the quadrate to the postorbital bar, the lower tem- poral fenestra would have been very small, especially with the quadratojugal inserted. This contrasts with the much larger fenes- trae of Orthosuchus and Eopncumatosuchus. Anterior to the foramen nervi facialis, only a small part of the side wall of the braincase is preserved. This includes the posterior part of the laterosphenoid and the anterior part of the prootic, which is recessed on its external surface to form the rostral pneu- matic sinus. The laterosphenoid lies between the quadrate and parietal at the antero-vcntral margin of the superior temporal fenestra. It is broken anterior and lateral to this, but the presence of a cotylus in the ventral surface of the postorbital indicates that the laterosphenoid had a dorsal knob of typical archosaurian design. Vertebrae Five vertebrae are preserved — one cervical, two dorsal, and two caudal. All of these, except for the posterior caudal, are platy- coelous (“amphicoelous” of some authors). In this respect the centra resemble those of other Triassic crocodilians. There are 18 OCCASIONAL PAPERS MUSEUM OF NATURAL HISTORY no pleurocoels. The neuro-central suture is fused in all vertebrae, another indication of the adult nature of the specimen. Measure- ments arc given in Table 1. The cervical vertebra (Fig. 4) is probably the 5th or 6th mem- ber of the presacra] series. There is a short cervical rib attached to the centrum at two points — on a short transverse process (diapo- physis) and at the anteroventral corner of the centrum (parapophy- sis). Hie “body” of the rib is directed anteroposteriorly as it is in Recent crocodilians. It is expanded anteriorly to form an articula- tion for the posterior extension of the preceding rib. The centrum above and below the parapophysis is excavated so that the para- pophysis is very prominent. The proportion of length to height for the centrum is 1.2; the proportion of length to width 1.2. The centrum bears a short keel on the midline of the ventral surface. The neural spine is narrow and posteriorly situated. The cervical Figure 4. — Lcsothosuchus charigi, A-C. Anterior views of cervical (pos- sibly 5th or 6th), dorsal (approximately 10th), and caudal (possibly 3rd or 4th) centra. D, Block of crushed elements: anterior dorsal vertebra (ant. dor. vert. ); dorsal armor; ribs; right scapula. Scale is 1 cm. TRIASSIC CROCODILIAN FROM SOUTHERN AFRICA 19 can be compared with that of Orthosiicluis, as figured by Nash (1975). Lesothosuchus differs from that genus by having a much lower neural arch and a relatively larger neural canal. The two dorsal vertebrae arc from the anterior and middle parts of the dorsal column. The anterior vertebra was badly crushed in a block containing part of a scapula, dorsal osteoderms, ribs and a caudal vertebra (Fig. 4). The anterior dorsal retains a ventral parapophysis on the centrum, but situated near the base of the pedicel. In all crocodilians the parapophysis migrates dor- sally along the centrum in the transitional series between “typical” cervical and dorsal vertebrae. After the first few dorsals the para- pophysis and diapophysis lie adjacent to each other on the trans- verse process, a condition found in the anterior dorsal of Lcsotho- suchus. Romer (1956) and Colbert (1952) considered this type of articulation to be a primitive archosaurian feature. Among archosaurs, I know it to be absent only in phytosaurs, theropod dinosaurs, and birds. In these taxa the parapophysis migrates dor- sally only as far as the pedicel of the neural arch. In Orthosuchus the first four dorsals retain a parapophysis which is at least partly on the centrum while Recent Alligator have a central parapophysis only on the first three dorsals. Judging from the position of this facet, the anterior dorsal of the type specimen of Lesothosuchus is probably the 9th, 10th, or 11th presacral. There is no ventral keel on the centrum of this anterior dorsal, a feature shared with Orthosuchus. The parapophysis is very large, much larger than that of the cervicals. The transverse process ex- tends directly outward and flares slightly to form an articulation for the rib tubercle. The spine is high, but relatively narrow. Orthosuchus differs from Lesothosuchus in having a much broader neural spine and a higher neural arch. When Lesothosuchus is compared with eusuchians and Ortho- suchus, , the more posterior dorsal vertebra appears to be from the middorsal region, approximately at the 19th presacral position. The centrum is somewhat elongate, with a length /height propor- tion of 1.5 and a length/ width proportion of 1.6. It is constricted in the midline. The parapophysis and diapophvsis are rather widely separated on the transverse process. The transverse process is much less extensive than those of eusuchian dorsals of compara- ble centrum width (Fig. 7). The height of the neural spine cannot be determined. This vertebra closely resembles that of comparable position in Orthosuchus. The two caudals are from the anterior and middle parts of the tail. The more anterior one corresponds to the third or fourth caudal. It is much larger than the anterior caudals of otherwise comparably sized eusuchians. The centrum is broadly flattened ventrallv, with large facets for the caudal chevrons. The rib is 20 OCCASIONAL PAPERS MUSEUM OF NATURAL HISTORY fused and posteriorly situated. The centrum is somewhat excavated laterally. The neural spine is incomplete, but vertically directed. Only a single, large chevron is preserved, and it also belongs to the anterior part of the caudal series. The more posterior caudal was removed from a block which contained the dorsal ostcoderms, dorsal ribs and the anterior dorsal vertebra. The centrum is biconvex, elongate, and ventrally flat- tened. The rib is posteriorly situated and prominent, but, relative to the anterior part of the caudal series, it has moved higher up on the pedicel of the arch. The neural spine is incompletely preserved. Pectoral appendage and girdle Of the shoulder girdle, only the dorsal end of the right scapula and the right coracoid are preserved. The end of the scapula is preserved in the block mentioned above (Fig. 4). It is expanded dorsally and constricts sharply to a narrow waist at the point of breakage. The shape of the coracoid is distinctively crocodilian (Fig. 5D). There is an expanded, “ procoraeoid” process, an elongate shaft with a distinct “waist,” and an expanded distal end. The coracoid compares well with Broom's (1904) figure of Notochcnnpsa and with Nash's (1971) figure of Orthosuchus, although the coracoid of Orthosuchus appears to have a less expanded distal foot and to be somewhat more massive. Lesothosuchus has a delicate coracoid which tapers to a thin edge anterior to the glenoid. This “pro- coracoid” area is piereed by a eoracoidal foramen. As in modern eusuchians, the proximal (humeral) end bears a flattened surface for articulation with the scapula. Because of the angle between this facet and the plane of the coracoid, the scapula would have been directed dorsally at an angle to the coracoid, as in Recent eusuchians. The glenoid facet on the posterior surface is directed backward into the coracoid (parasagittal) plane. This contrasts with cu- suchians and mesosuehians, which have an outwardly directed glenoid fossa (Fig. 5E). A parasagittal orientation of the glenoid fossa would have only allowed a stance in which the humerus was held close to the body, rather than directed outward in a sprawling position. This type of glenoid was once thought to be unique to the dinosaurs among archosaurs (Bakker and Galton, 1974). It is now known to be present in orthosuchids, Protosuchus , Pseudhes- perosuchus , Sphenosuchus and the thecodontian, “ Mandasuchus .” In none of these taxa is the direction of the glenoid demonstrably different from that of a “fully erect” dinosaur, such as Ilypsilo- phodon. The planes of orientation used in the following descrip- TRIASSIC CROCODILIAN FROM SOUTHERN AFRICA 21 tion presume a parasagittal and nearly vertical orientation for the limbs, the primitively dorsal surface of the humerus being rotated to a posterior, and possibly somewhat lateral position; that of the femur to a medial position. Lcsothosuchus lacks a “biceps” tubercle on the lateral surface of the coracoid. Such a tubercle is present on coracoids of some theropods (Ostrom, 1976) and Sphenosuchus (Walker, 1972). The structure previously called the biceps tubercle in Archaeopteryx is probably the attachment for the furcula (Whetstone, Ms.). In the position of the tubercle, which probably served as the origin of the supracoracoideus muscle, the Lesotho specimen has a rugose Figure 5. — A-D, Lcsothosuchus charigi, BM (NH) R 8503: A-C, restora- tions of the left humerus at natural size; D, lateral view of right coracoid, x2; E, Alligator, juvenile coracoid, X^. 22 OCCASIONAL PAPERS MUSEUM OF NATURAL HISTORY ridge just anterior to the glenoid and sternal to the coracoid fora- men. Eusuchians have a similar rugosity. Of the forelimb, the left and right humeri and a left ulna were preserved. The humerus is long and slender, with somewhat less torsion than in eusuchians (Fig. 5A-C). It is difficult to compare the humeral torsion with Orthosuchus , since Nash’s restorations of that genus seem to have the distal end incorrectly placed. Lcso- thosuchus has an expanded humeral head that is convexly rounded and oval in cross section. The medial surface is set at a slight angle. Orthosuchus and eusuchians have a more flattened and less enlarged humeral head. Lcsothosuchus has a long deltoid crest, though incompletely preserved on both humeri of the type speci- men. The posterior surface of the distal extremity is more de- pressed than in eusuchians. The supracondylar ridges are pro- nounced. The distal articular surface is composed of two rounded condyles, separated by a groove. In Recent eusuchians the humeral condyles are well differentiated as part of a series of adaptations in the elbow which result in the distal movement of the radius, relative to the ulna, during forearm flexion (see Walker, 1972, 1974). Lcsothosuchus lacks this condylar differentation. The ulna is slender, with compressed proximal and distal ends (Fig. 7C). It lacks the medial facet on the head that is critical to the operation of the “push-rod” mechanism discussed above. The proximal end is very unusual in having an egg-shaped knob on the lateral surface for articulation with the radius. Tin’s would have allowed an extensive rotation surface for that bone. Nothing com- parable is known among other archosaurs. The proximal articular surface consists of a depression on either side of the lateral knob. These depressions fit the condyles of the humerus. Orthosuchus and eusuchians differ from Lcsothosuchus in having poorly formed humero-ulnar articulations. By articulating the ulna with both humeral condyles, Lcsothosuchus has shifted the olecranon medi- ally under at least part of the medial condyle. It is unlikely, how- ever, that the ulna excluded the radius from that condyle since this would preclude pronation in a conventional humeral orienta- tion. The shaft of the ulna is slender, slightly curved and somewhat twisted. On the anterior surface of the distal end is a depressed, rugose insertion for the flexor carpi ulnaris. Just lateral to this ru- gosity is an elongate facet, the articulation for the medial corner of the radiale. This articulation is longer and more depressed than the corresponding structure in Recent crocodilians. Pelvic appendage The shaft and distal end of the left femur and the distal end of the right femur were recovered (Fig. 6G-J). The right shaft is broken just below the “fourth” trochanter. Judging from general TRIASSIC CROCODILIAN FROM SOUTHERN AFRICA 23 „ ... '..W ' “ A B Figure 6 . — Lcsothosuchus charigi, BM (NH) R 8503: A, B, medial and lateral views of left dentary; C, D, posterior aspects of right and left humeri; E, proximal articular surface of left ulna; F, dorsal osteoderm; G-I, left femur in medial, posterior, and lateral aspects; J, portion of right femur; K, L, right and left second metatarsals. Scale is 1 cm. 24 OCCASIONAL PAPERS MUSEUM OF NATURAL HISTORY proportions found in other crocodilians, the length of the femur was at least 70 mm. The shaft is slightly built and strongly re- curved antero-posteriorly. The dorsal surface of the distal end bears two epicondylar ridges. The lateral and medial surfaces on either side of the epicondylar ridges are flattened. The medial sur- face bears a pronounced groove for the medial collateral liga- ments. There is a great disparity between the tibial and fibular condyles. The tibial condyle is more rounded, extends farther dis- tally and is lower dorso-ventrally. There is a deep intercondylar notch on the posterior surface with little articular surface connect- ing the two condyles. A similar morphology is present in Ortho- suchtts. The fibular shaft is very slender and is sigmoidally curved in the antero-posterior plane. The proximal end is compressed. The distal end is slightly crushed, but appears to have been more com- pressed than in eusuchians. The fibula of Orthosuchus is only fig- ured in anterior and posterior aspects by Nash but it is shorter and much more massive than that of Lesothosuclius (Fig. 7). Little can be said of the metatarsals (Fig. 6) and phalanges. They do not differ significantly from the corresponding elements in Orthosuchus or eusuchians. Ribs and Osteoderms Two dorsal ribs are preserved. One is an anterior rib with broadly separated head and tubercle. It is unusual in the degree of dorso-vcntral compression of the vertebral processes. The shaft expands distally, but is incomplete. The more posterior rib is long and slender. The proximal end is not adequately exposed for description. The ribs were associated with four dorsal osteoderms, pre- served so that none is completely exposed (Fig. 4, 6F). Two ap- pear to be complete, but are crushed into each other so that the dorsal surface is not visible. The outline of these bones resembles the dorsal armor of Orthosuchus . The plates are rectangular, slightly curved dorso-ventrally and have an anterior peg, pre- sumably to underlie the preceding plate. There is no indication of a lateral flange, but this may be the result of breakage. One incomplete scute has the dorsal surface exposed. The sculpture of fine pits and ridges is similar to that on the skull roof. This sculp- ture is quite distinct from that on the dorsal osteoderms of Ortho- suchus, which has large, deep pits. There is a thickened, unsculp- tured ridge on the leading edge of the plate, which marks the overlap of the preceding plate. Four smaller plates are presumably from the caudal region. They lack anterior processes, but have a thickened leading edge. A curious element of uncertain position has a size and sculpture TRIAS SIC CROCODILIAN FROM SOUTHERN AFRICA 25 Figure 7. — Mid-dorsal vertebra, left ulna (left figure) and right fibula (right figure) of A. juvenile Alligator ; B. Orthosuchus stonnhergi (after Nash 1975); C. Lcsothosuclius charigi ; all X 1 to show the proportions of meso- podials to the width of the dorsal centrum. The arrow indicates the position of the rounded articulation for the radius. 26 OCCASIONAL PAPERS MUSEUM OF NATURAL HISTORY similar to the small osteoderms, but is a thickened wedge. It could be interpreted as part of the ventral fold armor under the limbs, but since the related Orthosuchus lacks ventral armor, the bone remains indeterminate. DISCUSSION Mode of Life A striking feature of the type specimen of Lesothosuchus is the long and slender build of the limbs. This is even more apparent when the limbs are compared with an extant eusuchian of com- parable body size. Because of the incomplete state of the vertebral column, it is not possible to express the relative length of the limbs in terms of presacral length. Instead, I have used the width of the 19th presacral centrum. Because of the weight-bearing function of the dorsal vertebrae, the vertebral width seems, in- tuitively, to be a satisfactory indicator of body size. Comparative indices for limbs of Lesothosuchus , Orthosuchus , and some Recent crocodylids are given in Table 2. Lesothosuchus is dramatically different in its limb proportions; even from the closely related Triassic Orthosuchus . This limb elongation raises questions re- garding the locomotor pattern of Lesothosuchus and other Triassic crocodilians. Modern crocodilians are able to assume a variety of stance and gait postures beyond the resting “sprawl” by which they have been stereotyped. When moving overland, the limbs are usually shifted into the parasagittal plane and the body lifted up from the ground. This “high walk” has been figured by Schaeffer (1941). Cott (1961) has described another method of locomotion whereby small crocodiles gallop or hop in the maimer of a squirrel. Still Table 2. — Limb lengths of some Triassic and Recent crocodilians expressed as a proportion of dorsal vertebral width. Lesothosuchus chart gi (type) Orthosuchus stormhergi ( type ) Stcgomosuchus longipcs ( typo ) Recent Crocodylidac Humerus 9 7 8 5-7 Ulna 8 7 6 4-6 Femur 11 9 9 6-9 Fibula 11 8 7 5-7 * Indices are limb lengths divided by the width of the 19th presacral centrum. In Lesothosuchus the exact position of the vertebra can not be known with certainty, but the width of the centrum varies little in this region. Data for Orthosuchus are from Nash (1975); data for Stcgomosuchus from Emerson and Loomis (1904). The presacral width is used in the indices for Stcgomosuchus. Indices of Recent crocodilians reflect the range of indices obtained on all taxa available in the collections of the British Museum (Natural History). TRIASSIC CROCODILIAN FROM SOUTHERN AFRICA 27 another possibility is full bipcdality (Cliarig, 1972). In recogni- tion of these varied locomotor postures and their improvement over the classic sprawl of lower tetrapods, Cliarig (1972) has termed the crocodyloid stance and gait “semi-improved.” By in- ference based upon skeletal comparisons, he further applied this terminology to the early crocodilians and pseudosuchians. Chang’ s approach to pseudosuchian locomotion was quite sound and long overdue. However, the analogy from modern forms may not be applicable to all of the Triassic crocodilians. Orthosuchus differs from eusuchians in almost every feature of the hindlimb which Cliarig (1972) used to separate the semi-im- proved and “erect” (parasagittal) archosaurs: the acetabulum is deep and perforate (though not fully open, it is as open as the bipedal Archaeopteryx); the supra-aeetabular crest is well devel- oped; there is ample origin for the protractor musculature on the anterior iliac crest (which extends considerably anterior to the acetabulum) and on the elongate, anteriorly directed pubis; and the femur has a well developed, medially directed head (though it lacks a distinct neck). Nash (1971) correspondingly restored Orthosuchus with an erect gait. Lesothosuchus was presumably similar to Orthosuchus in the morphology of the femur and hip, but differs from Orthosuchus and Recent crocodilians in other aspects of limb morphology. The limbs are relatively more slender and elongate, as discussed above. The humeral head is expanded and convex, increasing the degree of movement possible at the shoulder. The deltopectoral crest is long, increasing the vertical backswing leverage (Bakker and Gal- ton, 1974). The ulna has eotyles for articulation with both humeral condyles, increasing the structural stability of the elbow, but de- creasing or eliminating ulnar rotation. There is a unique condyle on the ulna for articulation with the radius, increasing the rotational capacity of the radius. The limbs arc greatly elongated and lightly built, more so in the mesopodium than the propodium, and more in the hind limb than the forelimb. Lesothosuchus resembles Orthosuchus but differs from extant eusuchians in having the glenoid directed along the plane of the body. From the morphology of the shoulder girdle and limbs, it seems likely that Lesothosuchus held its limbs close to the body, in the parasagittal plane, as do most Recent mammals. All of these features would seem to be specializations for cur- sorial progression. This progression could have been by running (cursoriality in the strict sense) or by the bounding gallop ob- served in living forms by Cott (1961). Walker (1970) has sug- gested that the latter mode of locomotion was primitive for the Croeodylia and that the customary sprawl of extant forms is a sec- ondary adaptation to an amphibious lifestyle. The differences be- 28 OCCASIONAL PAPERS MUSEUM OF NATURAL HISTORY tween the limb girdles of protosuchians and higher eroeodilians do indicate that their method of locomotion was quite different. Lull (1953) argued that Stegomosuchus , with proportionally shorter limbs than Lesothosuchus, was a small cursorial predator with para- sagittal limb posture. Although this may be the adaptive zone into which the earliest eroeodilians radiated, it is necessary to reconcile this mode of life with the cranium, whose morphology indicates the presence of an operculum (Ohrklappe) and an oblique tym- panum. In recent taxa these have generally been interpreted as adaptations for swimming. Considering the graeile build of spheno- suchids, the primitive “outgroup” for comparison with Triassie eroeodilians, it is difficult to accept an aquatic mode of life for the earliest eroeodilians. It seems likely that the otic structures must have evolved as a result of other factors. It is possible that the oblique tympanum was originally a consequence of akinesis, while the external ear functioned for sound reception, as it does in mammals. Relationships of Lesothosuchus Lesothosuchus resembles Stegomosuchus in the shape of the superior temporal fenestrae, but differs greatly in the proportions of the limbs to the skull roof and vertebral centra, and the relative size of the dorsal armor. The type specimen of Stegomosuchus preserves few other features for comparison. Lesothosuchus differs from Orthosuchus in many features: the shape and size of the temporal fenestrae, the structure of the occiput, lateral margin of the squamosal, cervical vertebrae, anterior dorsal vertebrae, cora- coid, humerus, ulna and fibula, the length of the limbs relative to the vertebrae and skull roof, and the sculpture of the dorsal armor. However, it shares with Orthosuchus the highly fenestrated quad- rate and the absence of a posterior notch on the dentary for the mandibular foramen, both of these features are unique to them among the Croeodylia (see Crompton and Smith, 1980, for Proto- suchtis). These two genera comprise the new subfamily Ortho- suehinae. Orthosuchines and Protosuchus share features presumed to be primitive for the Croeodylia, despite spotty distributions among the comparative outgroups: laterally directed orbits, superficial postorbital bar, two palpebral bones, primary palate, loose posterior quadrate articulation, platyeoelous centra, anterior iliac crest, supra- aeetabular ridge, and elongate, rod-like pubis. These characteristics partly diagnose the grade, “Protosuehia.” With eusuehians and mesosuehians, Lesothosuchus shares ad- vanced skeletal features which are not present, as a group, in pseudosuehians or proterosuehians: the postorbital bone antero- posteriorly reduced; the parietals fused; the supratemporal fossa TRIASSIC CROCODILIAN FROM SOUTHERN AFRICA 29 anteromedially placed so that it lies over, rather than alongside the brainease (enlargement of the fossa in mesosuchians is pre- sumed secondary); the lateral wall of the inside of that fossa formed partly by the quadrate and laterosphenoid; the descending process of the squamosal lost; the quadrate strongly inclined, with two articular faces, sutured to the prootic, laterosphenoid, squa- mosal and postorbital, and pneumatic; the otic notch “closed” by the posterior articulation of the quadrate; periotic pneumatic cavi- ties developed; the squamosal forming a shelf over the otic re- gion; the posttemporal fenestra reduced; the supraoccipital ex- cluded from the foramen magnum; the internal carotid artery enclosed by bone posterior to its entrance into the otic region; the coracoid elongate, expanded sternally, and with a distinct waist. Features unknown for Lesothosuchus , but which are probably an- cestral for the Crocodylia, include: anterior supraorbital bone present; loss of serrations on the teeth; closure of the basal articu- lation in the skull by suture; enclosure of the eustachian tubes in the bones of the basieranium; loss of clavicles; elongation of the proximal carpals; and presence of a calcaneal heel. These features can be added to those above to complete my diagnosis of the “Protosuchia.” Relationships of Sphenosuchus Sphenosuchus and its allies remain a problematical group which seems to be closely related to crocodilians. Their morphology can be variously interpreted as support for at least three phylogenetic hypotheses. Reversing his earlier opinion, Walker (1974) argued that Sphe- nosuchus was closer to birds than to crocodilians. The features that he used can be categorized as: 1) being absent in Spheno- suchus (e.g. streptostylic-kinetic system), 2) being common to ether “higher” archosaurs (e.g. elongate cochlear recess), 3) being absent in the earliest birds (e.g. elongate coracoids), or 4) being present in both birds and true crocodilians. Sphenosuchus does, however, share with birds the contribution of the prootic bone to the posterior quadrate cotyle, a feature which may be absent in crocodilians. Walker (1970) had originally placed Sphenosuchus closer to crocodilians. This relationship is supported by the presence of elongated proximal carpal bones in Sphenosuchus , Pseudhespero - suchus , and the undescribcd genus from the Welsh fissure deposits and the elongation of the dorsal process of the quadratojugal in Sphenosuchus . These characteristics are otherwise unique to croco- dilians. This phylogeny could also be supported by the “akinetic” skull of Sphenosuchus, an interpretation contrary to Walker’s (1974). Metakinensis, the presumed primitive condition in archo- 30 OCCASIONAL PAPERS MUSEUM OF NATURAL HISTORY saurs, is precluded in Sphenosuchus by the broad suturing of the occipital elements. Streptostyly, the avian condition, is not possi- ble in Sphenosuchus since the alinement of the two heads of the quadrate in the plane of the cheek and the elongation of the an- terior head of the quadrate would prevent the quadrate from rock- ing in its articulation. The anterior head forms a “stop” against streptostyly. Streptostyly is also prevented by the unbroken post- orbital bar. Sphenosuchus has the jugal and postorbital bones tightly sutured within the bar. Moreover, their contact is too vertical to allow them to slide apart with an appreciable antero- posterior vector, even if the contact were loose. The snout is held tightly together by squamous sutures which would prevent sliding as envisioned by Walker (1974). While these features preclude an avian-type intracranial joint, some slight movements may have been possible at the dorsal end of the quadrate, which retains ball and socket joints. Because of the substantial differences in the morphology of the quadrate and quadrate articulations between Sphenosuchus and croeodilians, it is difficult to evaluate the significance of akinesis as a functional similarity. There is also some doubt as to the primi- tive condition of the cranium in birds, which have generally been presumed to have retained an ancestral kinesis by shifting the joint progressively forward (Bock, 1964). Although Bock (1964) predicted that the braincase of Archaeopteryx would be meso- kinetic, new preparation of the London specimen of Archaeopteryx (Whetstone, 1983) shows that it is not. “Akinesis” as a character unifying Sphenosuchus and croeodilians is the result of a single shared morphology, the broad, tight suturing of the skull roof to the more posterior and ventral occipital elements, a feature which is also shared with birds. A third phylogenetic hypothesis, which I support, as detailed below, is that birds and croeodilians share a unique crocodylo- morph ancestry, with Sphenosuchus as a sister group. Although accepting this phylogenv requires that one accept the loss of elon- gate carpals in the avian stem, the weight of the cranial evidence may justify it. Origin of birds Walker (1972, 1974) was the first to suggest that birds and croeodilians have a common ancestry independent of dinosaurs. His argument was based upon comparisons between Sphenosuchus and Recent birds. There is now additional evidence to support a common-ancestor hypothesis in the cranial morphology of the earliest croeodilians and of Mesozoic and Recent birds. The de- tailed morphology of these birds and arguments regarding avian character polarity are presented in a separate paper (Whetstone, TRIASSIC CROCODILIAN FROM SOUTHERN AFRICA 31 MS.). Lcsothosuchus shares many derived cranial features with primitive birds: a hollow quadrate bone; periotic pneumatic sinuses in the dorsal, central and rostral positions; a bipartite quadrate articulation with the braincase; an anterior articulation formed, at least in part, by the laterosphcnoid, squamosal and prootic; and a posterior cotyle for the quadrate at the anterior base of the paroeciput and formed, at least in part, by the otoccipi- tal. The structure and position of the paroccipital process may also be a shared derived feature at this level, but the character distribution is such that it is difficult to determine the primitive situation in birds. The detailed similarity in the construction of these features and their distribution in fossil archosaurs and birds strongly supports their homologous origin in an ancestor unique to both crocodilians and birds. This hypothesis is contrary to the popularly accepted theory of avian descent from carnivorous dinosaurs (Ostrom, 1976), a theory supported primarily by the structure of the carpus and maims in Archaeopteryx. The quadrate is hollow in most Recent birds, including the flightless ratites. This hollow cavity has a medial foramen connect- ing with the middle ear, an independent feature which is also shared with crocodilians, but which is probably convergcntly de- rived within the Avcs. Among Mesozoic birds, the quadrates of Archaeopteryx and Gobipteryx appear to be hollow, while those of the hesperornithids appear to be solid, a characteristic of many heavy-boned diving birds (e.g. penguins). Only Ichthyornis is known to have a medial pneumatic foramen. To the best of my knowledge a hollow quadrate bone lias never been described in a dinosaur, sphcnosuchid or thecodontian and none of these have medial pneumatic foramina. The primitive position of the antrum pneumaticum dorsale in crocodilians is between the supraoccipital, otoccipital, squamosal and parietal in that part of the cranium where all of these elements interconnect. The sinus connects with the middle ear just dorsal to the facies articularis posterior via a foramen at or near the junc- tion of the squamosal, prootic and otoccipital. The sinus and its lateral foramen have this structure in juveniles of most Recent birds and in the Mesozoic hesperornithiforms. Sphcnosuchus has a dorsal sinus but the details of its construction are as yet unde- scribed. This sinus is absent in both saurischian and ornithischian dinosaurs and in thecodontians. The antrum pneumaticum centrale of crocodilians is located in the base of the paroccipital process. In Lcsothosuchus it under- lies and medially undercuts the posterior cotylus for the quadrate. In Recent eusuchians the opening into the middle ear cavity is just ventral to the posterior quadrate articulation and immediately pos- terior to the fenestra ovalis. In most Recent birds the sinus has 32 OCCASIONAL PAPERS MUSEUM OF NATURAL HISTORY this same morphology, except that the foramen is bordered partly by the prootic. It is present in Hesperornis, and probably in Enaliornis. Only one dinosaur is known to develop a similar struc- ture, the ornithomimid, Gallimimus. In that genus the paroccipital process is hollow and has two anterior foramina (see figures 5 and 6 by Osmolska et al. , 1972, where the foramina are misidentified as the fenestra ovalis and fenestra rotunda). This cavity does not underlie or undercut the quadrate articulation, which is a single cup in the squamosal situated near the lateral terminus of the paroccipital process. This morphology is apparently a convergence to that of birds or crocodilians. The theropods, AUosaurus, Droniae- osaurus, and Tyrannosaurus all lack the cavity, as do sphenosuchids and thecodontians. The antrum pneumaticum rostrale of early crocodilians, such as Eopneumatosuchus, is situated just anterior to the foramen for the facial nerve and posterior to the trigeminal foramen. It is overlain by an anterior process from the quadrate, and overlies the carotid and eustachian canals. This is also true of most modern birds, with the notable exception of penguins, which lack the cavity. Birds differ from crocodilians in that the lateral wall of the sinus is formed by the alasphenoid part of the parasphenoid, presumably an autapomorphy for birds. The sinus is present in hesperornithiforms, including the Lower Cretaceous Enaliornis. An antrum pneumaticum rostrale is absent in theocodontians and theropod dinosaurs, although the ornithischian Ilypsilophodon has a depression in the position of the sinus. Among the archosaurs, the quadrate articulation is bipartite only in sphenosuchids, crocodilians and birds. With the exception of the highly sutured phytosaurs, thecodontians and dinosaurs have a single socket for the quadrate which is formed by the squamosal bone. There is no contact with either the laterosphenoid or prootic. The squamosal extends laterally as far as the terminus of the paroccipital process and separates the quadrate from the otoeeipital. Essentially the same morphology is found in Sphenodon. The na- ture of the quadrate articulations in primitive birds have generally been misunderstood. This is because of the unfortunate descrip- tion of the quadrate as ‘one-headed’ and because of the fusion of otic elements in the adult skull. The facies artieularis anterior of primitive crocodilians is formed by the pr ootic, squamosal, laterosphenoid and postorbital. In ratites and the hesperornithids, it is formed by the prootic, squamosal and laterosphenoid. This is presumed to be the primitive situation for birds, although the laterosphenoid component is lost in most neognathous taxa. Birds and crocodilians are the only archosaurs that articulate the quadrate with the prootic and laterosphenoid. TRIASSIC CROCODILIAN FROM SOUTHERN AFRICA 33 In Sphenosuchus the anterior articulation is formed only by the squamosal bone. Primitive birds ( Archaeopteryx , hesperornithids, ratites, and possibly penguins) resemble Lesotho su cl uis in having the facies articularis posterior as a cotyle situated at the anterior base of the paroccipital process. It is formed, at least in part, by the otoccipital. This cotyle is medially underlain and undercut by the central periotic sinus. New preparation of the London specimen of Archaeopteryx shows this articulation to be very similar to that of Lesothosuchus. In Ilesperornis and many Recent birds, there is a canal just medial to the cotyle which connects the dorsal and central periotic pneumatic cavities. Because Sphenosuchus lacks many of the unique homologies of birds and crocodilians, I consider it to be more remotely related to both groups. Sphenosuchus has a dorsal periotic sinus, but lacks central and rostral pneumatic sinuses. The quadrate articulation is bipartite as in birds and crocodilians. The posterior articulation is a cotyle, which is partly formed by the prootic, but not the otoccipital. The anterior articulation is formed by the squamosal only. The anterior articulation resembles that of birds in being a cotyle, but this is the primitive form of the primary (“squamosal”) articulation in arehosaurs. Martin, Stewart and Whetstone (1980) have shown that the dentitions of crocodilians and Mesozoic birds are unique among arehosaurs in having an unserrated crown with a constricted waist, and an expanded root with an oval, enclosed resorption pit. Sphenosuchus has a primitive, thecodontian-like dentition, with serrations, no waist, and straight roots. If one argues that birds and crocodilians are only remotely related, it would be necessary to interpret each of these features as convergence. This seems unlikely in light of the detailed struc- ture which is shared and the differences in mode of life between crocodilians and birds. When the alternative hypotheses are com- pared, the considerable cranial evidence favors a sister-group rela- tionship between birds, crocodilians and sphenosuchids (Fig. 8). ACKNOWLEDGEMENTS For allowing me to examine specimens, I thank A. J. Charig, C. A. Walker, G. Cowles, N. Arnold and P. Crush (London); G. Viohl (Eichstatt); H. Jaeger and H. Fischer (Berlin); S. L. Olson and N. Hotton (Washington); A. D. Walker (Newcastle); E. S. Gaffney (New York); J. Ostrom (New Plaven); W. Turnbull and J. Bolt (Chicago); C. L. Forbes (Cambridge, England); B. J. Pyrah (Yorkshire); P. Calton (Bridgeport); A. Elzanowski and II. Osmolska (Warsaw) and L. D. Martin, W Duellman, R. Men gel, and M. J. Mengel (Kansas). I must particularly thank Dr. and Mrs. A. D. Walker for allowing me to stay at their home while in 34 OCCASIONAL PAPERS MUSEUM OF NATURAL HISTORY Newcastle. M. A. Klotz and C. D. Whetstone prepared some of the figures. A. D. Walker, E. O. Wiley, H. P. Schultze, and R. M. Mengel read the manuscript and made many helpful suggestions. Advice and encouragement from L. D. Martin and A. J. Charig were particularly appreciated. Funding was provided by NSF DEB 7821432, Sigma Xi, F. M. Chapman Fund, a Fulbright-Hays Fellowship, the National Geographic Society, and a Dissertation Fellowship from the University of Kansas. A. J. Charig, P. J. Whybrow, John Attrige, and lone Rudner were members of the field party which collected the type specimen. BIRDS Figure 8 . — Phylogenetic hypothesis advocated by the author. Character suites in the skull which are discussed in the text are 1 ) antrum pneumaticum dorsale, bipartite quadrate articulation; 2) prootic and laterosphenoid con- tributing to anterior quadrate articulation, otoccipital contributing to posterior quadrate articulation, antrum pneumaticum rostrale and centrale, quadrate pneumaticity; 3) quadrate sutured to the braincase, anterior quadrate articu- lation extended dorsally, posterior portion of the carotid enclosed, prootic contribution to the posterior articulation lost (not known with certainty in Lesothosuchm ), basipterygoid articulation sutured (inferred in Lcsothosuchus by comparison with Orthosuchus). TRIASSIC CROCODILIAN FROM SOUTHERN AFRICA 35 LITERATURE CITED Bakker, R. and P. Galton 1974. Dinosaur monophyly and a new class of vertebrates. Nature, 248: 168-172. Bock, W. 1964. Kinetics of the avian skull. J. Morph., 114:1-42. Broili, F. and J. Schroeder 1936. Beobachtungen an Erythrochampsa Haughton. Sber. bayer. Akad. Wiss., 1936:229-238. Broom, R. 1904. On a new crocodilian genus (N oto champs a) from the upper Storm- berg beds of South Africa. Geol. Mag., 1:582-584. Brown, B. 1933. An ancestral crocodile. American Mus. Novitates., 638:1-4. 1934. A change of names. Science, 79:80. Buffetaut, E. 1979. The evolution of the crocodilians. Sci. American, 241:130-144. Charig, A. J. 1969 Expeditions: Lesotho. In Report on the British Museum (Natural History) 1966-1968, British Museum (Natural History), London, pp. 31-35. 1972. The evolution of the archosaur pelvis and hind-limb: an explana- tion in functional terms. In Studies in vertebrate evolution (Ed. K. A. Joysey and T. S. Kemp), Winchester Press, New York, pp. 121-155. Colbert, E. H. 1952. A pseudosuchian reptile from Arizona. Bull. American Mus. Nat. Hist., 99:565-592. Colbert, E. H. and C. C. Mook 1951. The ancestral crocodilian Protosuchus. Bull. American Mus. Nat. Hist., 97:143-182. Cott, H. B. 1961. Scientific results of an enquiry into the ecology and economic status of the Nile crocodile in Uganda and Northern Rhodesia. Trans. Zool. Soc. London, 29:211-356. Crompton, A. W. and K. Smith 1980. A new genus and species of crocodilian from the Kayenta forma- tion (Late Triassic ?) of Northern Arizona. In Aspects of verte- brate history (Ed. L. Jacobs), Museum of Northern Arizona Press, Flagstaff, pp. 193-217. Crow, T. M. and A. A. Crow 1979. Anatomy of the vascular system of the head and neck of the helmeted guinea fowl Nnmida meleagris. J. Zook, London, 188: 221-233. Emerson, B. K. and F. B. Loomis 1904. On S tegomus longipes , a new reptile from the Triassic sandstones of the Connecticut Valley. American Sci., ser. 4( 17 ) :377-380. Haughton, S. II. 1924. The fauna and stratigraphy of the Stormberg series. Ann. S. African Mus., 12:323-497. Hennig, W. 1966. Phylogenetic systematics, Univ. Illinois Press, Urbana. Ilochstetter, F. 1906. Beitrage zur Anatomie und Entwickelungsgeschichte des BlutgefaB- systemes der Krokodile. In Reise in Ostafrika in den Jahren 1903- 1905, Stuttgart, 1-139 pp. 36 OCCASIONAL PAPERS MUSEUM OF NATURAL HISTORY Huenc, F. von 1922. The Triassic order Thecodontia. American J. Sci., ser. 5(4):22-26. 1925. Die Bedeutung der Sphenosuchus — Gruppe fur den Ursprung der Krokodile. Z. indukt. Abstamm. Vererblehre, 38:307-322. Iordansky, N. N. 1973. The skull of the crocodilia. In Biology of the Reptilia (Ed. C. Gans and T. S. Parsons), Academic Press, New York, vol. 4, pp. 201 - 262 . Kermack, K. 1956. An ancestral crocodile from South Wales. Proc. Linn. Soc. London, 166:1-2. Krebs, B. 1976. Pseudosuchia. In Handbuch der Paliioherpetologie 13 (Ed. O. Kuhn), Gustav Fischer Verlag, Stuttgart, pp. 40-98. Lull, R. S. 1953. Triassic life of the Connecticut valley. Bull. Geol. Survey Con- necticut, 81:1-336. Martin, L. D., J. D. Stewart and K. N. Whetstone 1980. The origin of birds: structure of the tarsus and teeth. Auk, 97:86-93. Nash, D. 1968. A crocodile from the Upper Triassic of Lesotho. J. Zool. London, 156:163-179. 1971. The morphology and relationships of a crocodilian, Orthosuchus stormbergi , from the Upper Triassic of Lesotho. Unpublished Ph.D. dissertation, University of London. 1975. The morphology and relationships of a crocodilian, Orthosuchus stormbergi, from the Upper Triassic of Lesotho. Ann. S. African Mus., 67:227-329. Osmolska, H., E. Roniewicz and R. Barsbold 1972. A new dinosaur, GaUimimus bullatus n. gen., n. sp. ( Ornithomimi- dae) from the Upper Cretaceous of Mongolia. Palaeontologia Polonica, 27:103-143. Ostrom, J. 1976. Archaeopteryx and the origin of birds. Biol. J. Linn. Soc., 8: 91-182. Patterson, C. and D. Rosen 1977. Review of ichthyodcctiform and other Mesozoic tclcost fishes and the theory and practice of classifying fossils. Bull. American Mus. Natur. Hist, 158:81-172. Romer, A. S. 1956. Osteology of the reptiles. Univ. Chicago Press, Chicago, 772 pp. Saiff, E. 1974. The middle car of the skull of birds, the Procellariiformes. Zool. J. Linn. Soc, 54:213-240. 1976. Anatomy of the middle ear region of the avian skull: Sphenisci- formes. Auk, 93:749-759. 1978. The middle ear of the skull of birds: the Pelecaniformes and Ciconiiformcs. Zool. J. Linn. Soc, 63:315-370. (in press). The middle car of the skull of birds: the Struthioniformes. Zool. J. Linn. Soc. Schaeffer, B. 1941. The morphological and functional evolution of the tarsus in am- phibians and reptiles. Bull. American Mus. Natur. Hist, 78:395- 472. TRIASSIC CROCODILIAN FROM SOUTHERN AFRICA 37 Simmons, D. J. 1965. The non-therapsid reptiles of the Lufeng Basin, Unnan, China. Fieldiana, Geol., 15:1-93. Walker, A. D. 1968. Protosuchus, Protcrocliampsa , and the origin of phytosaurs and crocodiles. Geol. Mag., 105:1-14. 1970. A revision of the Jurassic reptile H allopus victor (Marsh), with re- marks on the classification of crocodiles. Phil. Trans. R. Soc., B257: 323-372. 1972. New light on the origin of birds and crocodiles. Nature, 237: 257-263. 1974. Evolution, organic. In Yearbook of Science and Technology 1974 (Ed. D. N. Lapedes), McGraw-Hill, New York, pp. 177-179. Whetstone, K. N. 1983. Braincase of Mesozoic birds: I. New preparation of the “London” Archaeopteryx. Jour. Vert. Paleont., 2:439-452. Whetstone, K. N. and L. D. Martin 1979. New look at the origin of birds and crocodiles. Nature, 279: 234-236. 1981. Common ancestry for birds and crocodiles? (Reply). Nature, 289:98.